Release rates under inert atmosphere

The ratios of tritium normalised release rates under inert atmosphere are presented in Figure 39. It can be seen that at all temperatures in inert atmosphere tritium evolved faster than ¹²C from the graphite matrix. The release rate ratio decreased with time and later became constant.

The high rate ratio at the beginning can be related to desorption of radionuclides from the

62

0 50 100 150 200 250 300

0 2 4 6 8 10 12 14 16

Time, hours NR3 H/NR12 C

MM 870°C, argon MP 970°C, argon MP 1060°C, argon MM 1060°C, argon AVR 1060°C, argon

10

Figure 39. Ratio of normalised release rate of ³H to normalised release rate of ¹²C in inert atmosphere

As has been established [80], hydrogen atoms are trapped at the edges of the graphene sheets and between the graphene layers near the particle surface. Two kinds of trapping sites exist for hydrogen retention [81–83]. One is an interstitial cluster loop edge or solitary dangling bond located in a crystallite with adsorption enthalpy of 4.4 eV. The other is a carbon dangling bond located at the edge surface of a crystallite with an enthalpy of 2.3 eV [83]

(Figure 40). In the case of neutron- or ion-irradiated graphite, the Trap 1 sites dominate hydrogen retention and Trap 2 are important in the case of unirradiated graphite.

Figure 40. Model of hydrogen trapping sites in a graphite material

Desorption of hydrogen starts at ~300°C and has two maxima around 450°C and 780°C [84].

The first peak is considered to be due to hydrogen desorbed from defective sites between graphite sheets, and the second one originates from the covalent bonding sites.

The release of tritium from graphite grains is controlled by the diffusion process. Tritium atoms desorb from the trapping sites. Then their association reaction takes place at the open

pores as well as on the inner and outer surface of polycrystalline particles, and tritium molecules can diffuse away through open pores to free space.

The mechanism of tritium diffusion in graphite is described in details in the literature.

Numerous studies have been performed on hydrogen diffusion and retention in graphite [41–

44, 81, 82, 84].

Most commercial graphites are not single crystalline materials, but consist of aggregates of polycrystalline particles. The microstructural model is presented in Figure 41. It is assumed [43] that the tritium atoms captured by carbon atoms diffuse in a crystalline grain along the a-and c-axis. In addition, tritium atoms can also diffuse along the grain boundaries [85]. Three distinct diffusion channels are indicated in Figure 41 by arrows: 1 – along the a-axis in a grain, 2 – along the c-axis in a grain and 3 – along the grain boundary. Diffusion through path 2 does not appear to be plausible because of the long-distance carbon network and hence considerable large activation energy.

Figure 41. Microstructural models of isotropic and anisotropic graphite Diffusion coefficients of tritium in graphite

Diffusion coefficients of tritium in graphite can be determined from tritium release curves during isothermal heating of contaminated graphite samples. The fractional release of tritium from graphite F(t) can be plotted as a function of the square root of the heating time, t(s). As can be seen from Figure 42, tritium release fits well with this function. The diffusion coefficient, D (cm²/ s^-1), was calculated from the slope of the linear portion of the curve using the following relation, which was derived from Fick’s law [86]:

( ) 4 Dt 1

F t S ˜R Eq. (62)

for F(t) d 0.3

64

R – the recoil range of tritium from ⁶Li(n,D)³H reaction in graphite. The recoil range for graphite was taken from the literature and amounted to 32 Pm [86, 87]. Results of the calculations are presented in Table 22.

Table 22. Diffusion coefficients of tritium in Merlin and AVR graphite

Type of graphite Temperature, °C Diffusion coefficient D, cm²/ s^-1

Merlin 870 5.70E-14

Merlin 970 7.40E-13

Merlin 1060 6.19E-12

AVR 1060 1.70E-12

0 0.2 0.4 0.6

0 10 20 30 40

Time^1/2, min^1/2

Fraction release (t)

MB 870°C MP 970°C MB 1060°C AVR 1060°C MP 1060°C

Figure 42. Tritium release vs. square root of the treatment time

Figure 43 shows an Arrhenius plot of logarithmic diffusion coefficients for Merlin graphite vs. 1/T (°K). The activation energy was calculated from the slope as 323 kJ/mol and the pre-exponential factor as 10.7 cm²/s.

The pre-exponential factor and activation energy are different for different types of graphite due to classification of the diffusion paths [43]. The diffusion of tritium in highly oriented graphite or highly anisotropic graphite will be slower than in isotropic graphite, because the tritium atoms hardly penetrate the layer plane. As was shown by Saeki [44] the activation energy of the diffusion process in graphite samples with a high value of the anisotropy factor was in the range of 250–260 kJ/mol, whereas the activation energy for isotropic pyrographite amounted to 105 kJ/mol. In the case of laminar graphite [87], with a quite large parameter l_a the activation energy is 412 kJ/mol for diffusion along grain boundaries. The values of pre-exponential factor and activation energy obtained for Merlin graphite in the present work are typical of nuclear graphite with a high anisotropy factor.

-37

Figure 43. Arrhenius plot for tritium diffusion in Merlin graphite

14C normalised release rates

The ratio of the carbon isotope release rate (K14C/12C) was lower than in the case with tritium (Chapter D.2.3.1, Figure 39 ). The highest K value was observed for Merlin graphite by treatment at 870°C in argon. As can be seen from Figure 44, the K value was close for Merlin graphite treatment at 1060°C and 870°C. A higher release rate of ¹⁴C in comparison to ¹²C means that this radionuclide is not distributed homogeneously in the graphite grains otherwise it would be released at the same rate as ¹²C. Therefore it can be supposed that the graphite surface near grain boundaries is enriched with ¹⁴C.

Information about ¹⁴C localisation in graphite was not found in the literature. It is known that

14C is an activation product originating mainly from nitrogen impurities in graphite matrix or from nitrogen adsorbed on the surface from the gas cooler [30]. The contribution of ¹³C and

17O are of less importance.

One possibility for the ¹⁴C position in a graphite crystal is adsorption on the edge planes of the graphite surface. Oxygen functional groups on carbon surfaces decompose to carbon oxides upon heating in an inert atmosphere [88]. The complexes yielding CO2 were shown to decompose typically over a range of temperatures starting at 200°C and exhibiting desorption maxima at 300°C and 600°C [89]. Carboxylic acids have been proposed to be predominantly responsible for the low-temperature peak. The high-temperature CO₂ evolution has been attributed to carboxylic anhydrides. Similarly, the CO-yielding groups have a CO evolution maximum at 900 and 1100 K. The total desorbed quantity of CO and CO₂ amounted to 19 Pmol/g for graphitised carbon fibres with a specific surface area of 6 m²/g. This corresponds to 0.023 % of mass loss. In the present experiments, mass loss varied in the range from 0.16 to 1.6 %. Thus this carbon cannot be related to the desorption of functional groups alone but also to slow oxidation caused by the presence of oxygen traces in the inert carrier gas. Moreover, desorption would take place in the initial period of time, whereas constant release of ¹⁴C from the sample was observed throughout the heating procedure.

66

Figure 44. Ratio of normalised release rate of ¹⁴C to normalised release rate of ¹²C in inert atmosphere

The other possibility of radiocarbon arrangement in graphite is intercalation between graphite layers or embedding in graphite lattice under neutron irradiation [90]. ¹³C also can be a source of ¹⁴C uniformly distributed in the graphite matrix. Being in the composition of graphite lattice, ¹⁴C can be removed only by oxidation because carbon diffusion in the basal plane is significant only above 2200°C [34].

A carbon atom displacement caused by neutron or ion irradiation results in crystallite dimensional changes [20]. Two types of damage centres are known – vacancies and interstitials. Interstitials cause crystallite growth perpendicular to the layer planes (c-axis direction) and coalescence of vacancies causes shrinkage parallel to the layer plane (a-axis direction). During thermal treatment annealing of the lattice defects occurs. Interstitials start to migrate to the edge surface or to reintegrate with vacancies [20, 91, 92]. If ¹⁴C is in the composition of interstitials it can be supposed that during heating in inert atmosphere diffusion of ¹⁴C from graphite can be attributed to diffusion of interstitials from the interlayer space to the edges of graphite planes with their subsequent desorption. But, as was already mentioned above, the presence of a constant concentration of CO and CO₂ of a few ppm in the carrier gas flow supposes the presence of oxygen in the system. Therefore the conditions were not absolutely “oxygen-free” and release of ¹⁴C caused by oxidation cannot be excluded.